A Moving Object Detection Method Based on Level Set in Dynamic Scenes

The article is to demonstrate a way to detect moving objects in dynamic scene. This method integrates the means of background difference and level set. First, we adopt background difference, which is to connect the color of video image and movement mode, and then we can renew the dynamic scene by judging the threshold so that the foreground contour initial model can be obtained. Next, the level set can be evolved, through which the more exact foreground objective can be detected. Because of the introduction of movement mode during the course of obtaining the background, the judgment of the background of dynamic scene is more precise. The experiment result suggests that, under the dynamic scene, like ripples and light, the means of level set is the best real-time way to detect the moving objects. However, it is unlikely to detect more than one object in dynamic scene, which is also the future research direction.